Common questions and quick fixes

Use the checks below before changing a project or starting another run.

​

I cannot find my project

Check the signed-in account, workspace selector, and project selector. If the workspace is missing, ask its administrator to check your access.

​

The result page is empty

Check whether a run has completed. Confirm the selected run, then clear narrow filters. An empty view can mean no data for that selection; it is not automatically a zero result.

​

The trend chart is missing

Trend analysis needs at least two completed runs. A project with one completed run can still show its current citations, mentions, and sentiment.

​

Two pages show different numbers

Compare the run, platform, topic, intent, prompt type, entity selection, and metric label. Citation share, citation coverage, raw citation counts, and mention coverage are different measures.

​

My brand is mentioned, but my website is not cited

A mention records the name appearing in an answer. A citation records a source reference. One does not require the other.

​

The run is complete, but Results stability says Not converged

Processing has finished, but the citation-ranking assessment has not passed both checks. Read Rank stability and Domain SNR together. Below the displayed SNR target of 1.00, the result lacks sufficient separation from noise; above the target, rank stability still matters. Treat close ranks as provisional, review the prompts and scope, and plan further measurement if needed. Read the explanation.

​

How many prompts are enough?

Use the setup count and quality warnings, then judge the completed result. No fixed count guarantees a decision-grade result for every topic and platform. The aim is sufficient signal and stable rankings, with confidence intervals precise enough for your decision. Collecting far beyond that point may use resources for little extra benefit.

​

The number changed. Should we act?

Check the relevant confidence interval and the assessment in Trend analysis. No significant change means the comparison has not established a clear change beyond uncertainty. It does not prove nothing changed. For a significant result, also consider the size of the difference and the supporting evidence.

​

Should my prompts include my brand?

Leave brand names, recognizable product names, and brand acronyms out of a non-branded discovery project. Use a separate branded project when you want to examine what AI says about you or which competitors it introduces into answers about your brand. See examples.

​

Does a domain capture every page on my website?

The domain identifies which captured pages belong to your website. Available pages depend on what the run found and could capture. Check Top cited URLs and Webpage analysis; for a missing page you want to edit, see the options under Content → Open existing content.

​

My uploaded file is rejected

Check the format and the 3 MB limit. Bring prompts accepts CSV, TXT, and XLSX. Bring queries accepts CSV, ZIP, and XLSX. Use the screen’s sample or template, review the row messages, and confirm that you chose the right route for questions versus short search phrases.

​

I want to change the topics after launch

The established configuration is locked. Use Create a new project from Project settings to measure different topics, brand terms, or pages.

​

I cannot find a content item

Open Content and clear Find content, Writing status, and Source filters. Check additional pages and the selected project. A document title can differ from its original recommendation title.

​

My draft changed, but its score did not

The visible score may belong to the previous text. Wait for the edit to save and choose Update score when offered.

​

I want the previous draft back

Open History, preview the version, and use the restore action if you want it to replace the working text. Read the confirmation first.

​

The draft says In review. Is it approved?

No. In review is a writing label. Your team still needs to complete its editorial and publishing process.

​

I exported content, but the recommendation is still Open

Exporting is a handoff step. Complete the real action first, then use Record implementation and describe what went live.

​

The recommendation has no result after implementation

A later completed run may be needed. Read the Results message and the comparison dates. Some outcomes cannot be compared; do not fill in missing values yourself.

​

A button is disabled

Read the nearby message. Common reasons include missing required information, unsaved edits, an active operation, role permissions, or plan limits. Resolve that specific reason before retrying.

​

Where can I find Agent Q?

Use Ask the agent on Overview, or the Agent button on supported project pages. The right-hand panel lets you ask beside your work. See the walkthrough.

​

Where did my Agent Q conversation go?

Check the selected project and conversation title. The docked panel shows conversations associated with the page. Use Expand panel, then Show chats, to browse or search conversations across the project. Starting a new conversation does not erase your saved discussions.

​

Can I use ChatGPT or Claude instead?

Yes. Open your account Settings → MCP → Connect and choose the supported setup option. Once connected, ask the assistant to list your IQRush projects, then select one. The connection guide explains setup, scope, and access.

​

Will my conversations move between assistants?

Agent Q, ChatGPT, and Claude have separate conversation histories. Carry over a short summary with the project, run, filters, and question when moving between them.

​

The assistant and the page show different results

Check the project, run date, filters, metric, and aggregation. Ask the assistant to identify its source and scope. If the difference remains, keep both readings and the context for your support contact; do not choose a number simply because it appears in a confident answer.

​

Can I request a report for an audience we did not measure?

Ask which audience profiles and prompts exist first. The assistant can help outline the questions for a new assessment; measured findings for that audience require relevant data. See the question and report gallery.

​

I need help explaining a problem

Record the page name, project, selected run, filters, and the exact message shown. A screenshot of the affected area can help your team or support contact understand the same state. Next: Use the glossary or return to Start here.